pith. sign in

arxiv: 2507.20185 · v2 · submitted 2025-07-27 · 💻 cs.CL

SessionIntentBench: A Multi-task Inter-session Intention-shift Modeling Benchmark for E-commerce Customer Behavior Understanding

Yuqi Yang , Weiqi Wang , Baixuan Xu , Wei Fan , Qing Zong , Chunkit Chan , Zheye Deng , Xin Liu
show 8 more authors
Yifan Gao Changlong Yu Chen Luo Yang Li Zheng Li Qingyu Yin Bing Yin Yangqiu Song
This is my paper

Pith reviewed 2026-05-19 02:51 UTC · model grok-4.3

classification 💻 cs.CL
keywords intention modelinge-commerce sessionsbenchmarkintention shiftlarge language modelscustomer behaviorsession analysismultimodal evaluation
0
0 comments X p. Extension

The pith

Current large vision-language models fail to track and use shifting customer intentions across e-commerce sessions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces an intention tree concept along with a curation pipeline to mine customer intentions from browsing session logs at scale. It builds SessionIntentBench, a multimodal benchmark containing four subtasks that test how well L(V)LMs handle inter-session intention shifts, using over a million intention entries and trajectories drawn from more than ten thousand sessions. Experiments on a human-annotated subset show that existing models cannot effectively capture or apply these intentions in complex settings, yet explicitly supplying intention information improves LLM results on the tasks. A sympathetic reader would care because accurate intention tracking could support more precise predictions of user preferences during online shopping.

Core claim

We introduce the concept of an intention tree and propose a dataset curation pipeline. Together, we construct a sibling multimodal benchmark, SessionIntentBench, that evaluates L(V)LMs' capability on understanding inter-session intention shift with four subtasks. With 1,952,177 intention entries, 1,132,145 session intention trajectories, and 13,003,664 available tasks mined using 10,905 sessions, we provide a scalable way to exploit the existing session data for customer intention understanding. Extensive experiments on the annotated data further confirm that current L(V)LMs fail to capture and utilize the intention across the complex session setting. Further analysis show injecting the 1,

What carries the argument

The intention tree, a hierarchical structure that extracts and organizes customer intentions from multi-product session logs to represent shifts between sessions and support evaluation on the four subtasks.

If this is right

  • Injecting intention data into LLMs improves performance on customer behavior understanding tasks.
  • The curation pipeline enables scalable exploitation of existing e-commerce session logs for intention modeling.
  • Human-annotated ground truth provides a reliable evaluation set for testing model capability on inter-session shifts.
  • Models that better capture intentions can more accurately predict on-the-spot user preferences from session features.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • E-commerce platforms could use intention trees to refine recommendations when users leave a product page due to unmet features.
  • The same hierarchical intention extraction method might transfer to other sequential user domains such as travel planning or content consumption.
  • Training future models to generate intention trees directly from raw logs could reduce reliance on post-hoc curation.

Load-bearing premise

The intention tree and dataset curation pipeline accurately extract and represent true customer intentions from session logs without introducing systematic bias or oversimplification.

What would settle it

Run the four subtasks on a new set of sessions where users explicitly report their intentions in follow-up surveys; if models given injected intention data show no performance gain over baselines, the central claim is falsified.

Figures

Figures reproduced from arXiv: 2507.20185 by Baixuan Xu, Bing Yin, Changlong Yu, Chen Luo, Chunkit Chan, Qingyu Yin, Qing Zong, Wei Fan, Weiqi Wang, Xin Liu, Yang Li, Yangqiu Song, Yifan Gao, Yuqi Yang, Zheng Li, Zheye Deng.

Figure 1
Figure 1. Figure 1: An example of customer intention-shift in the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of SESSIONINTENTBENCH and the construction pipeline. Multi-modal attribute extraction is conducted first as an aid for further step intention generation. Metadata analyses are conducted afterward to provide a more fine-grained and detailed inspection of intention shifts in the session interaction. Here, Ai , Ci stands for attributes and comparisons at i-th step. Different task is associated with d… view at source ↗
Figure 3
Figure 3. Figure 3: Radar chart of models that have the best per [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison between best performances across different methods on different tasks. All base￾line’s max accuracy line is consistent with the max ac￾curacy line over open models of all methods (i.e., Zero￾shot, fine-tuning with SESSIONINTENTBENCH (SIB), and sequential fine-tuning with MIND then SIB). Pay￾ing for proprietary API does not add extra value in this case. 5.4 The Impact of Intention Injection Obser… view at source ↗
read the original abstract

Session history is a common way of recording user interacting behaviors throughout a browsing activity with multiple products. For example, if an user clicks a product webpage and then leaves, it might because there are certain features that don't satisfy the user, which serve as an important indicator of on-the-spot user preferences. However, all prior works fail to capture and model customer intention effectively because insufficient information exploitation and only apparent information like descriptions and titles are used. There is also a lack of data and corresponding benchmark for explicitly modeling intention in E-commerce product purchase sessions. To address these issues, we introduce the concept of an intention tree and propose a dataset curation pipeline. Together, we construct a sibling multimodal benchmark, SessionIntentBench, that evaluates L(V)LMs' capability on understanding inter-session intention shift with four subtasks. With 1,952,177 intention entries, 1,132,145 session intention trajectories, and 13,003,664 available tasks mined using 10,905 sessions, we provide a scalable way to exploit the existing session data for customer intention understanding. We conduct human annotations to collect ground-truth label for a subset of collected data to form an evaluation gold set. Extensive experiments on the annotated data further confirm that current L(V)LMs fail to capture and utilize the intention across the complex session setting. Further analysis show injecting intention enhances LLMs' performances.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces SessionIntentBench, a multimodal benchmark for evaluating L(V)LMs on inter-session intention-shift modeling in e-commerce customer behavior. It defines an 'intention tree' concept and an automated curation pipeline to mine 1,952,177 intention entries and 1,132,145 trajectories from 10,905 sessions, yielding four subtasks and over 13 million tasks. Human annotations create a gold evaluation set; experiments on this set show current L(V)LMs fail to capture and utilize intentions across sessions, while explicit intention injection improves performance.

Significance. If the benchmark construction is shown to be faithful, the work supplies a scalable, large-scale resource for intention-aware session modeling that could drive progress in e-commerce recommendation and user-behavior understanding. The empirical demonstration of model limitations and the benefit of intention injection offers concrete, falsifiable guidance for future LVLM development in this domain.

major comments (2)
  1. [Dataset Curation Pipeline] Dataset curation pipeline (described in the methods section following the abstract): the intention tree extraction rules and automated pipeline are load-bearing for all central claims, yet no independent validation against direct user-reported intent, surveys, or external signals is reported. Without this, it remains possible that observed LVLM failures and injection benefits are artifacts of systematic oversimplification (e.g., privileging explicit clicks over implicit feature dissatisfaction).
  2. [Experiments] Experiments section (on the annotated gold set): the manuscript states that experiments 'further confirm' model failures and injection benefits, but provides insufficient detail on error analysis, controls for post-hoc data selection, or sensitivity to curation hyperparameters. This weakens the ability to rule out that results depend on choices made after seeing initial model outputs.
minor comments (2)
  1. [Abstract] The term 'sibling multimodal benchmark' in the abstract is unclear; replace with a precise description of the modality and task structure.
  2. [Throughout] Notation for models alternates between L(V)LMs, LVLMs, and LLMs; standardize throughout and define the scope of 'V' explicitly.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback. We address each major comment below, providing clarifications on our approach and indicating revisions where appropriate to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Dataset Curation Pipeline] Dataset curation pipeline (described in the methods section following the abstract): the intention tree extraction rules and automated pipeline are load-bearing for all central claims, yet no independent validation against direct user-reported intent, surveys, or external signals is reported. Without this, it remains possible that observed LVLM failures and injection benefits are artifacts of systematic oversimplification (e.g., privileging explicit clicks over implicit feature dissatisfaction).

    Authors: We acknowledge the value of external validation signals. The automated pipeline extracts intention trees from observable session behaviors (clicks, views, dwell time, and navigation sequences) that serve as established proxies for implicit user intent in e-commerce literature. Direct user surveys at the scale of nearly 2 million entries are not feasible within the scope of this benchmark construction. However, the human-annotated gold set provides an independent, expert-verified check on the extracted intentions and trajectories. We will add a dedicated limitations subsection discussing potential biases in the automated rules and how the gold-set annotations help mitigate concerns about oversimplification. revision: partial

  2. Referee: [Experiments] Experiments section (on the annotated gold set): the manuscript states that experiments 'further confirm' model failures and injection benefits, but provides insufficient detail on error analysis, controls for post-hoc data selection, or sensitivity to curation hyperparameters. This weakens the ability to rule out that results depend on choices made after seeing initial model outputs.

    Authors: We agree that greater transparency is needed. In the revised manuscript we will expand the experiments section with (1) a detailed error analysis categorized by subtask, session length, and intention-shift type; (2) explicit controls comparing performance on the annotated gold set versus randomly held-out subsets; and (3) sensitivity results across variations in key curation hyperparameters (intention-tree depth, filtering thresholds, and trajectory length). These additions will demonstrate that the reported model limitations and injection benefits are robust to the curation choices. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical benchmark construction with external human validation

full rationale

The paper proposes an intention tree concept and curation pipeline to mine SessionIntentBench from 10,905 sessions, yielding millions of entries. It explicitly conducts human annotations to create a ground-truth gold set for evaluation. All claims about L(V)LMs failing to capture inter-session intention shifts and the benefit of intention injection are supported by direct model evaluations on this annotated data. No equations, fitted parameters, or derivations are present that reduce to the paper's own inputs by construction. The work is self-contained as an empirical benchmark contribution rather than a closed logical loop.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claims rest on the assumption that session logs contain extractable intention signals and that the proposed tree structure faithfully represents them. No free parameters or invented physical entities are described.

axioms (1)
  • domain assumption Session history logs contain sufficient implicit signals to infer customer intentions that change across sessions.
    Invoked in the motivation for the intention tree and curation pipeline.
invented entities (1)
  • intention tree no independent evidence
    purpose: To organize and represent hierarchical customer intentions from session data.
    New concept introduced to address insufficient information exploitation in prior works.

pith-pipeline@v0.9.0 · 5828 in / 1288 out tokens · 23912 ms · 2026-05-19T02:51:26.872031+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages · 10 internal anchors

  1. [1]

    Bruce L Alford and Abhijit Biswas. 2002. https://doi.org/10.1016/S0148-2963(00)00214-9 The effects of discount level, price consciousness and sale proneness on consumers' price perception and behavioral intention . Journal of Business Research, 55(9):775--783

    work page doi:10.1016/s0148-2963(00)00214-9 2002

  2. [2]

    Ebtesam Almazrouei, Hamza Alobeidli, Abdulaziz Alshamsi, Alessandro Cappelli, Ruxandra Cojocaru, Mérouane Debbah, Étienne Goffinet, Daniel Hesslow, Julien Launay, Quentin Malartic, Daniele Mazzotta, Badreddine Noune, Baptiste Pannier, and Guilherme Penedo. 2023. https://arxiv.org/abs/2311.16867 The falcon series of open language models . Preprint, arXiv:2...

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  3. [3]

    Miguel Alves Gomes, Richard Meyes, Philipp Meisen, and Tobias Meisen. 2022. https://doi.org/10.1145/3511808.3557127 Will this online shopping session succeed? predicting customer's purchase intention using embeddings . In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, CIKM '22, page 2873–2882, New York, NY, USA...

    work page doi:10.1145/3511808.3557127 2022

  4. [4]

    Diddigi Raghu Ram Bharadwaj, Lakshya Kumar, Saif Jawaid, and Sreekanth Vempati. 2022. https://arxiv.org/abs/2210.15451 Fine-grained session recommendations in e-commerce using deep reinforcement learning . Preprint, arXiv:2210.15451

    work page arXiv 2022

  5. [5]

    Minjin Choi, Hye-young Kim, Hyunsouk Cho, and Jongwuk Lee. 2024. https://doi.org/10.1145/3626772.3657928 Multi-intent-aware session-based recommendation . In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR '24, page 2532–2536, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/3626772.3657928 2024

  6. [6]

    Honghua (Kathy) Dai, Lingzhi Zhao, Zaiqing Nie, Ji-Rong Wen, Lee Wang, and Ying Li. 2006. https://doi.org/10.1145/1135777.1135902 Detecting online commercial intention (oci) . In Proceedings of the 15th International Conference on World Wide Web, WWW '06, page 829–837, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/1135777.1135902 2006

  7. [7]

    Wenxuan Ding, Weiqi Wang, Sze Heng Douglas Kwok, Minghao Liu, Tianqing Fang, Jiaxin Bai, Xin Liu, Changlong Yu, Zheng Li, Chen Luo, Qingyu Yin, Bing Yin, Junxian He, and Yangqiu Song. 2024. https://doi.org/10.18653/v1/2024.findings-emnlp.123 I ntention QA : A benchmark for evaluating purchase intention comprehension abilities of language models in E -comm...

    work page doi:10.18653/v1/2024.findings-emnlp.123 2024

  8. [8]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao, Aston Zhang, Aurelien Rodriguez, Austen Gregerson, Ava S...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  9. [9]

    Yupeng Hou, Jiacheng Li, Zhankui He, An Yan, Xiusi Chen, and Julian McAuley. 2024. https://arxiv.org/abs/2403.03952 Bridging language and items for retrieval and recommendation . Preprint, arXiv:2403.03952

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  10. [10]

    Derek Hao Hu, Qiang Yang, and Ying Li. 2008. https://doi.org/10.1145/1517472.1517476 An algorithm for analyzing personalized online commercial intention . In Proceedings of the 2nd International Workshop on Data Mining and Audience Intelligence for Advertising, ADKDD '08, page 27–36, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/1517472.1517476 2008

  11. [11]

    Ravi Chandra Jammalamadaka, Naren Chittar, and Sanjay Ghatare. 2009. https://doi.org/10.1145/1620432.1620468 Mining product intention rules from transaction logs of an ecommerce portal . In Proceedings of the 2009 International Database Engineering & Applications Symposium, IDEAS '09, page 311–314, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/1620432.1620468 2009

  12. [12]

    Bohan Jia, Jian Cao, Shiyou Qian, Nengjun Zhu, Xin Dong, Liang Zhang, Lei Cheng, and Linjian Mo. 2023. https://doi.org/10.1145/3587099 Smone: A session-based recommendation model based on neighbor sessions with similar probabilistic intentions . ACM Trans. Knowl. Discov. Data, 17(8)

    work page doi:10.1145/3587099 2023

  13. [13]

    Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, Lélio Renard Lavaud, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, and William El Sayed. 2023. https://arxiv.org/abs/2310.0...

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  14. [14]

    Di Jin, Luzhi Wang, Yizhen Zheng, Guojie Song, Fei Jiang, Xiang Li, Wei Lin, and Shirui Pan. 2023 a . https://doi.org/10.1145/3543507.3583526 Dual intent enhanced graph neural network for session-based new item recommendation . In Proceedings of the ACM Web Conference 2023, WWW '23, page 684–693, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/3543507.3583526 2023

  15. [15]

    Wei Jin, Haitao Mao, Zheng Li, Haoming Jiang, Chen Luo, Hongzhi Wen, Haoyu Han, Hanqing Lu, Zhengyang Wang, Ruirui Li, Zhen Li, Monica Xiao Cheng, Rahul Goutam, Haiyang Zhang, Karthik Subbian, Suhang Wang, Yizhou Sun, Jiliang Tang, Bing Yin, and Xianfeng Tang. 2023 b . http://papers.nips.cc/paper\_files/paper/2023/hash/193df57a2366d032fb18dcac0698d09a-Abs...

    work page 2023

  16. [16]

    Lei Li, Yongfeng Zhang, and Li Chen. 2020. https://doi.org/10.1145/3340531.3411992 Generate neural template explanations for recommendation . In Proceedings of the 29th ACM International Conference on Information & Knowledge Management, CIKM '20, page 755–764, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/3340531.3411992 2020

  17. [17]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023 a . https://arxiv.org/abs/2304.08485 Visual instruction tuning . Preprint, arXiv:2304.08485

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  18. [18]

    Xin Liu, Zheng Li, Yifan Gao, Jingfeng Yang, Tianyu Cao, Zhengyang Wang, Bing Yin, and Yangqiu Song. 2023 b . https://arxiv.org/abs/2312.16199 Enhancing user intent capture in session-based recommendation with attribute patterns . Preprint, arXiv:2312.16199

    work page arXiv 2023

  19. [19]

    OpenAI. 2024 a . https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/ Gpt-4o mini: advancing cost-efficient intelligence . OpenAI

    work page 2024

  20. [20]

    OpenAI. 2024 b . https://openai.com/o1/ Introducing openai o1 . OpenAI

    work page 2024

  21. [21]

    OpenAI. 2025. https://openai.com/index/openai-o3-mini/ Openai o3-mini . OpenAI

    work page 2025

  22. [22]

    OpenAI, Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, Red Avila, Igor Babuschkin, Suchir Balaji, Valerie Balcom, Paul Baltescu, Haiming Bao, Mohammad Bavarian, Jeff Belgum, Irwan Bello, Jake Berdine, Gabriel Bernadett-Shapiro, Christopher Berner...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  23. [23]

    Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Mei Li, Mingfeng Xue, Pei Zhang, Qin Zhu, Rui Men, Runji Lin, Tianhao Li,...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  24. [24]

    Soyeon Shim, Mary Ann Eastlick, Sherry L Lotz, and Patricia Warrington. 2001. https://doi.org/10.1016/S0022-4359(01)00051-3 An online prepurchase intentions model: The role of intention to search . Journal of Retailing, 77(3):397--416

    work page doi:10.1016/s0022-4359(01)00051-3 2001

  25. [25]

    Zhu Sun, Hongyang Liu, Xinghua Qu, Kaidong Feng, Yan Wang, and Yew Soon Ong. 2024. https://doi.org/10.1145/3626772.3657688 Large language models for intent-driven session recommendations . In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR '24, page 324–334, New York, NY, USA. Associat...

    work page doi:10.1145/3626772.3657688 2024

  26. [26]

    Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, Léonard Hussenot, Thomas Mesnard, Bobak Shahriari, Alexandre Ramé, Johan Ferret, Peter Liu, Pouya Tafti, Abe Friesen, Michelle Casbon, Sabela Ramos, Ravin Kumar, Charline Le Lan, Sammy Jerome, Anton Tsitsulin, Nino Vieillard, Piotr Stanczyk, Sertan Girgin, ...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  27. [27]

    Dongjing Wang, Haojiang Yao, Dongjin Yu, Shiyu Song, He Weng, Guandong Xu, and Shuiguang Deng. 2025 a . https://doi.org/10.1016/j.neunet.2024.107062 Graph intention embedding neural network for tag-aware recommendation . Neural Networks, 184:107062

    work page doi:10.1016/j.neunet.2024.107062 2025

  28. [28]

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Yang Fan, Kai Dang, Mengfei Du, Xuancheng Ren, Rui Men, Dayiheng Liu, Chang Zhou, Jingren Zhou, and Junyang Lin. 2024 a . https://arxiv.org/abs/2409.12191 Qwen2-vl: Enhancing vision-language model's perception of the world at any resolutio...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  29. [29]

    Weiqi Wang, Limeng Cui, Xin Liu, Sreyashi Nag, Wenju Xu, Chen Luo, Sheikh Muhammad Sarwar, Yang Li, Hansu Gu, Hui Liu, Changlong Yu, Jiaxin Bai, Yifan Gao, Haiyang Zhang, Qi He, Shuiwang Ji, and Yangqiu Song. 2025 b . https://aclanthology.org/2025.acl-long.1/ Ecomscriptbench: A multi-task benchmark for e-commerce script planning via step-wise intention-dr...

    work page 2025

  30. [30]

    Weiqi Wang, Tianqing Fang, Wenxuan Ding, Baixuan Xu, Xin Liu, Yangqiu Song, and Antoine Bosselut. 2023 a . https://doi.org/10.18653/V1/2023.FINDINGS-EMNLP.902 CAR: conceptualization-augmented reasoner for zero-shot commonsense question answering . In Findings of the Association for Computational Linguistics: EMNLP 2023, Singapore, December 6-10, 2023 , pa...

    work page doi:10.18653/v1/2023.findings-emnlp.902 2023

  31. [31]

    Weiqi Wang, Tianqing Fang, Chunyang Li, Haochen Shi, Wenxuan Ding, Baixuan Xu, Zhaowei Wang, Jiaxin Bai, Xin Liu, Cheng Jiayang, Chunkit Chan, and Yangqiu Song. 2024 b . https://doi.org/10.18653/V1/2024.ACL-LONG.128 CANDLE: iterative conceptualization and instantiation distillation from large language models for commonsense reasoning . In Proceedings of t...

    work page doi:10.18653/v1/2024.acl-long.128 2024

  32. [32]

    Weiqi Wang, Tianqing Fang, Baixuan Xu, Chun Yi Louis Bo, Yangqiu Song, and Lei Chen. 2023 b . https://doi.org/10.18653/V1/2023.ACL-LONG.733 CAT: A contextualized conceptualization and instantiation framework for commonsense reasoning . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2...

    work page doi:10.18653/v1/2023.acl-long.733 2023

  33. [33]

    Weiqi Wang and Yangqiu Song. 2025. https://aclanthology.org/2025.acl-long.79/ MARS: benchmarking the metaphysical reasoning abilities of language models with a multi-task evaluation dataset . In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2025, Vienna, Austria, July 27 - August 1, 20...

    work page 2025

  34. [34]

    Yu Wang, Amin Javari, Janani Balaji, Walid Shalaby, Tyler Derr, and Xiquan Cui. 2024 c . https://doi.org/10.1145/3589335.3648324 Knowledge graph-based session recommendation with session-adaptive propagation . In Companion Proceedings of the ACM Web Conference 2024, WWW '24, page 264–273, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/3589335.3648324 2024

  35. [35]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, and Denny Zhou. 2023. https://arxiv.org/abs/2201.11903 Chain-of-thought prompting elicits reasoning in large language models . Preprint, arXiv:2201.11903

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  36. [36]

    Baixuan Xu, Weiqi Wang, Haochen Shi, Wenxuan Ding, Huihao Jing, Tianqing Fang, Jiaxin Bai, Xin Liu, Changlong Yu, Zheng Li, Chen Luo, Qingyu Yin, Bing Yin, Long Chen, and Yangqiu Song. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.446 MIND : Multimodal shopping intention distillation from large vision-language models for E -commerce purchase understan...

    work page doi:10.18653/v1/2024.emnlp-main.446 2024

  37. [37]

    Changlong Yu, Weiqi Wang, Xin Liu, Jiaxin Bai, Yangqiu Song, Zheng Li, Yifan Gao, Tianyu Cao, and Bing Yin. 2023. https://arxiv.org/abs/2211.08316 Folkscope: Intention knowledge graph construction for e-commerce commonsense discovery . Preprint, arXiv:2211.08316

    work page arXiv 2023

  38. [38]

    Chenwei Zhang, Wei Fan, Nan Du, and Philip S. Yu. 2016. https://doi.org/10.1145/2872427.2874810 Mining user intentions from medical queries: A neural network based heterogeneous jointly modeling approach . In Proceedings of the 25th International Conference on World Wide Web, WWW '16, page 1373–1384, Republic and Canton of Geneva, CHE. International World...

    work page doi:10.1145/2872427.2874810 2016

  39. [39]

    Qi Zhao, Yi Zhang, Daniel Friedman, and Fangfang Tan. 2015. https://doi.org/10.1145/2792838.2800178 E-commerce recommendation with personalized promotion . In Proceedings of the 9th ACM Conference on Recommender Systems, RecSys '15, page 219–226, New York, NY, USA. Association for Computing Machinery

    work page doi:10.1145/2792838.2800178 2015

  40. [40]

    Xi Zhu, Fake Lin, Ziwei Zhao, Tong Xu, Xiangyu Zhao, Zikai Yin, Xueying Li, and Enhong Chen. 2024. https://doi.org/10.1145/3696417 Multi-behavior recommendation with personalized directed acyclic behavior graphs . ACM Trans. Inf. Syst., 43(1)

    work page doi:10.1145/3696417 2024

  41. [41]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

    work page

  42. [42]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

    work page